Primary instruction source for AI agents. This file is project-owned — stack, commands, architecture, and conventions live here. The kit-owned methodology rules are imported under
## Methodology, so a methodology upgrade never touches this file. Keep it updated as the project evolves.
| Layer | Tech |
|---|---|
| Shell / Platform | {TAURI or ELECTRON or WAILS or WEB or CLI or MOBILE or LIBRARY} |
| Frontend | {REACT or VUE or SVELTE or ANGULAR or SOLID or HTMX or VANILLA or NONE} |
| Backend | {RUST or NODE or PYTHON or GO or CSHARP or JAVA or ELIXIR or NONE} |
| Database | {SQLITE or POSTGRES or MYSQL or MONGODB or SURREALDB or NONE} |
| Auth | {LOGIN_METHOD} |
| Language | {LANGUAGE} |
| Command | Where | Description |
|---|---|---|
{DEV_COMMAND} |
root | Start dev server |
{BUILD_COMMAND} |
root | Build for production |
{TEST_COMMAND} |
root | Run tests |
{LINT_COMMAND} |
root | Lint code |
{FRONTEND_COMPONENT}
→ {API_CALL}({endpoint}, {args})
→ {BACKEND_HANDLER}
→ {SERVICE}
→ {REPOSITORY} (local data)
→ {ADAPTER} (external API)
- Routing: {ROUTING_STRATEGY}
- Auth: {AUTH_MECHANISM}
Full conventions in .specs/memory/conventions.md. Summary:
- Frontend: {STYLING_APPROACH}, {UI_LIBRARY}, state via {STATE_MANAGEMENT}
- Backend: Repository (data), Adapter (external API), Service (business logic), {HANDLER_PATTERN}, naming: {NAMING_CONVENTION}
- New endpoint pattern: model → repository → service → handler → register → add type definition
- New adapter pattern: for external services, create adapter → inject into service → test with mock
- Client-side filtering: memoized/computed filter from input state
- IDs: {ID_STRATEGY}
- Clean Code: SOLID principles in
.specs/memory/clean-code.md
| Framework | {TEST_FRAMEWORK} |
| Coverage | {COVERAGE_TOOL} (threshold: {COVERAGE_THRESHOLD}%) |
| Run tests | {TEST_COMMAND} |
| Watch mode | {TEST_WATCH_COMMAND} |
| Coverage report | {TEST_COVERAGE_COMMAND} |
Full testing conventions in .specs/memory/conventions.md## Testing. TDD is mandatory (tests
before implementation) — see the methodology rules below.
This project follows the spec-driven + TDD methodology. Its operational rules — Key Rules, the change path, skills, and the memory/consistency model — are kit-owned and imported here, so methodology upgrades replace one file and never touch this one. Read and follow them; the import inlines the content for harnesses that support it (Claude Code), and names the file for those that don't:
@.specs/methodology.md